another language reconstruction question

 It seems that the language diversity of the world was far greater long ago
that it is now. Territories having language unity today had often much more
languages before, as atested by regions we know well, like europe, or USA, or
to what we can see when we look at the languages people use primitive
civilisations (aborigens for example).
 So what is the meaning in finding the common root of IE words as they were
spoken in proto IE, since proto IE was probably a collection of hundred times
more languages than the IE family has today. I dont understand how the accurate
reconstruction of "the original PIE root", can be accurate at all, since it
probably refers to hundred or thousands of maybe related, yet different words.
 PIE is not so far back in time and it already seems strange to me, but what
about attemps on reconstructing languages much older, like Nostratic, or
Eurasiatic? It seems nonsense to me.

Reply